The complete SDD plugin that keeps your requirements, code, and tests in sync. 13 agents, 57 MCP tools, 22 prompts, 14 hooks, 10-phase enforced pipeline, six compliance frameworks.
One install, everything works.
Created by Paula Silva · @paulasilvatech · @paulanunes85
Specky running inside VS Code · GitHub Copilot Agent Mode · EARS notation, enforced pipeline
Encapsulates multiple related components for distribution and reuse across environments. One install, one version lock, one rollback path.
Plugins are the enterprise best practice because they package complete solutions for distribution and reuse. Instead of each developer improvising, the team shares a governed, versioned, and auditable development workflow.
Encapsulate approved commands and workflows. Security policies restrict what the AI agent can do, ensuring only homologated APIs and tools are used.
Senior engineers codify project expertise into skills and hooks. New team members get the full context automatically — tribal knowledge lives in the repo, not in people's heads.
Built on the open Model Context Protocol standard. Connect to databases, Slack, Jira, GitHub, Azure DevOps — no vendor lock-in, real-time context from every source.
Versioned with a lock file. Test, update, and roll back independently. apm install ensures every team member runs the exact same configuration.
Specialized AI personas — @specky-orchestrator coordinates the full pipeline, @specky-onboarding guides setup, plus 11 more for every phase.
Install via APM (Agent Package Manager) from Microsoft. Install APM first, then apm install paulasilvatech/specky — 13 agents, 22 prompts, 8 skills, 14 hooks.
13 agents orchestrate the pipeline, 22 prompts for every phase, 8 skills with domain knowledge, 14 hooks guard transitions. One install.
HIPAA, SOC 2, GDPR, PCI DSS, ISO 27001. Validate specs against regulatory controls before writing code.
Zero outbound calls. Two dependencies. No eval(). Works fully offline and in air-gapped environments.
Push to GitHub, Terraform, Docker, Jira, Azure DevOps, Figma. Specky generates payloads, your client delivers.
Six patterns that guarantee every requirement is testable, traceable, and unambiguous.
Artifact validation, branch checks, phase gates, LGTM gates, security scan, spec sync, drift monitor. Pre and post on every phase.
One command installs the MCP server, 13 agents, 22 prompts, 8 skills, and 14 hooks.
# Install APM (one-time) — pick one: $ curl -sSL https://aka.ms/apm-unix | sh # macOS / Linux $ brew install microsoft/apm/apm # Homebrew $ irm https://aka.ms/apm-windows | iex # Windows PowerShell # Then install Specky: $ apm install paulasilvatech/specky
Everything is auto-configured. APM creates .vscode/mcp.json, installs agents, prompts, skills, and hooks automatically.
Open Copilot Chat → Agent mode → type @specky-onboarding, @specky-orchestrator, @sdd-init, @implementer, or use prompts like /specky-onboarding.
In Copilot Chat (Agent mode), type:
@sdd-init initialize spec for user authentication
Tip: Specky auto-detects your stack (TypeScript, Python, Java, Go…) and applies the right EARS patterns. No configuration needed for most projects. Full docs →
One command installs the MCP server, agents, prompts, skills, and hooks for Claude Code.
# Install APM (one-time) — pick one: $ curl -sSL https://aka.ms/apm-unix | sh # macOS / Linux $ brew install microsoft/apm/apm # Homebrew $ irm https://aka.ms/apm-windows | iex # Windows PowerShell # Then install Specky: $ apm install paulasilvatech/specky
Everything is auto-configured. APM sets up the MCP server and slash commands automatically.
Specky adds 22 prompts. Start with:
/sdd:spec write a spec for user auth /sdd:design design the auth architecture /sdd:review check traceability
All 12 commands: /sdd:spec /sdd:design /sdd:tasks /sdd:implement /sdd:test /sdd:review /sdd:comply /sdd:diagram /sdd:sync /sdd:gate /sdd:changelog /sdd:report
One command installs the MCP server, agents, prompts, skills, and hooks for Cursor and Windsurf.
# Install APM (one-time) — pick one: $ curl -sSL https://aka.ms/apm-unix | sh # macOS / Linux $ brew install microsoft/apm/apm # Homebrew $ irm https://aka.ms/apm-windows | iex # Windows PowerShell # Then install Specky: $ apm install paulasilvatech/specky
Everything is auto-configured. APM sets up the MCP server automatically. Works with any MCP-compatible IDE.
Ask your AI assistant: "Use specky to write a spec for [feature]" or call tools directly from the MCP panel.
One command installs the MCP server, agents, prompts, skills, and hooks for VS Code.
# Install APM (one-time) — pick one: $ curl -sSL https://aka.ms/apm-unix | sh # macOS / Linux $ brew install microsoft/apm/apm # Homebrew $ irm https://aka.ms/apm-windows | iex # Windows PowerShell # Then install Specky: $ apm install paulasilvatech/specky
Everything is auto-configured. APM creates .vscode/mcp.json automatically. Requires VS Code 1.99+ with GitHub Copilot extension.
Open Copilot Chat → Agent mode → type @specky list tools to confirm 57 tools are registered.
Air-gapped mode: Specky makes zero outbound network calls. It runs entirely on localhost via stdio. No telemetry, no cloud sync, no external dependencies beyond Node.js.
$ npm pack specky-sdd # produces: specky-sdd-3.3.0.tgz
$ npm install -g ./specky-sdd-3.3.0.tgz
The stdio transport works identically on air-gapped machines. Use the same mcp.json config as any other environment.
HIPAA, SOC 2, GDPR, PCI DSS, ISO 27001 validation runs locally against bundled rule sets — no external validation service required.
Enterprise security features (RBAC, rate limiting, HMAC-SHA256 integrity, audit log with syslog export) are available as of v3.3.0. See Security docs →
What is APM? APM (Agent Package Manager) is Microsoft's open-source dependency manager for AI agent configuration. It manages agents, skills, prompts, hooks, and MCP servers as versioned packages with lock files. Install APM first, then run apm install paulasilvatech/specky.
Auto-scan codebase, detect stack, map existing patterns
Reasoning modelEARS notation requirements — 6 patterns, fully testable
GPT-4o / SonnetArchitecture, components, interfaces, 17 diagram types
o3 / OpusTraceability matrix — every requirement linked to design
GPT-4o / SonnetSequenced task list with [P] parallel markers & gates
o3 / OpusCode generation guided by spec, SRP enforced per file
GPT-4o / SonnetTests for 6 frameworks — unit, integration, e2e, property
GPT-4o / SonnetOWASP Top 10 scan, compliance framework validation
o3 / OpusAPI docs, ADR, changelog, runbooks auto-generated
GPT-4o / SonnetBranch-aware PR (spec→develop→stage→main), blocking gates, sign-off
Reasoning model13 agents + 22 prompts
22 prompts + hooks
57 tools via MCP
stdio + HTTP transport
| Feature | Specky | Kiro (AWS) | Cursor | Windsurf | Antigravity |
|---|---|---|---|---|---|
| Architecture & Deployment | |||||
| What it is | SDD Plugin | IDE (VS Code fork) | IDE (VS Code fork) | IDE (VS Code fork) | IDE (Cloud+Desktop) |
| Runs 100% locally | ✓ Yes | ~ IDE local, AI cloud | ~ IDE local, AI cloud | ~ IDE local, AI cloud | ✗ Cloud-processed |
| Air-gapped / offline | ✓ Full | ✗ No | ~ Enterprise only | ~ Self-hosted | ✗ No |
| Open source (MIT) | ✓ MIT | ✗ Proprietary | ✗ Proprietary | ✗ Proprietary | ✗ Proprietary |
| Works in any IDE | ✓ Any MCP host | ✗ Kiro only | ✗ Cursor only | ✗ Windsurf only | ✗ Antigravity only |
| Spec-Driven Workflow | |||||
| Enforced pipeline phases | ✓ 10 phases | ~ 3 phases | ✗ None | ✗ None | ✗ None |
| EARS notation requirements | ✓ 6 patterns | ~ Basic EARS | ✗ No | ✗ No | ✗ No |
| Traceability matrix | ✓ Automated | ~ Manual | ✗ No | ✗ No | ✗ No |
| Spec-sync hooks | ✓ 14 hooks | ~ 3 event types | ✗ No | ✗ No | ✗ No |
| Autonomous agents | ✓ 13 agents | ~ 1 agent | ~ 1 (Composer) | ~ 1 (Cascade) | ~ 16 agents |
| Compliance & Security | |||||
| Compliance frameworks | ✓ 6 built-in | ✗ No | ✗ No | ~ SOC2/HIPAA | ✗ No certs |
| OWASP scan in pipeline | ✓ Phase 08 | ✗ No | ~ MCP add-on | ✗ No | ✗ No |
| RBAC + audit log | ✓ Built-in | ~ AWS IAM | ~ Enterprise | ~ Enterprise | ✗ No |
| IDE & Tooling | |||||
| MCP tools | ✓ 57 tools | ~ MCP support | ~ MCP marketplace | ~ MCP support | ~ 1,500+ catalog |
| Diagram generation | ✓ 17 types | ~ Sequence only | ✗ No native | ✗ No native | ~ Visual verify |
| IaC generation | ✓ Terraform, Bicep, Docker | ✗ No | ~ Prompt-based | ~ Prompt-based | ~ Prompt-based |
| Pricing | |||||
| Cost | ✓ Free · MIT | Free–$200/mo | Free–$200/mo | Free–$60/mo | Free–$250/mo |
Fork it, extend it, audit it. No vendor lock, no seat pricing.
Not suggestions. Actual gates that prevent skipping steps.
No proprietary IDE. Use VS Code, Copilot, Claude Code, Cursor, or any MCP-compatible client.
Code never leaves your machine. Air-gapped environments fully supported.
Sequence, class, ERD, state, C4, deployment and more from your specs.
Vitest, Jest, Playwright, Pytest, JUnit, xUnit. Plus property-based testing.
Terraform, Bicep, Dockerfiles generated from your design docs.
HIPAA, SOC 2, GDPR, PCI DSS, ISO 27001. Validate before you code.
Opt-in RBAC, token-bucket rate limiting, HMAC-SHA256 state integrity, hash-chained audit log with syslog export.